RTA Draft Assistant

How it works

1. Simulated battles. We simulate a very large number of RTA battles with a simulator that follows the game's combat rules. Heroes fight with real community builds from the Fribbels Hero Library, at maximum progression.
2. A battle model. A model trained on those battles estimates which team wins a matchup, for each warfare rule.
3. Draft search. During your draft, the helper looks ahead over the remaining picks and bans for both players, assuming the enemy answers well, and ranks your options by the model's estimate.
4. Estimates, not guarantees. Real games are far less predictable than simulations, so the chances you see are scaled to how real RTA matches turned out. They are still estimates with community builds, not measured win rates: your gear, your play and the enemy's real choices change the result.

Your data

• Stays in your browser. Your hero pool, chosen builds and backups are kept only in this browser. Importing a Fribbels file happens entirely in your browser; the file is never uploaded.
• Sent only to answer a search. To compute recommendations, the draft so far, your pool, your chosen builds and the warfare rule are sent to the server. They are used for the search and kept only in the server's memory; they are never written to disk or to logs.
• Anonymous feedback. If you use “How did the battle go?”, we collect your win or loss, your optional comment and a summary of the draft with the helper's estimate. There are no accounts and we never ask who you are, so please keep personal details out of comments. Feedback is used only to improve the model.
• No ads or trackers. Fonts are loaded from Google Fonts, and a Cloudflare Turnstile check keeps bots out. Like most websites, the servers keep standard request logs (such as IP address, browser and time) for up to 30 days, and count requests per IP address in memory only, to prevent abuse.